Shape Optimization Designs Of Transmission Tower Based On Genetic Algorithm And Neural Network

The transmission line construction cost is extraordinarily expensiveat present due to tremendous steel consumption of transmission tower. Theexpense of transportation, construction and foundation also increaseswith the tower weight. So how to make the tower structural arrangementmore reasonable and the transmission tower lighter by means of the bestuse of material has become a big concern for electric power departmentand design department.The BP-GA (Error Back-propagation Algorithm and Genetic Algorithm)algorithm is put forward in this paper, which associates Genetic Algorithmwith BP artificial neural network (NN). The genetic algorithm is good atprocessing the optimized variables, which are in different condition,continuous or separate or continuous & separate. This method optimizesthe structural member cross-section area and node coordinates. To get thefitness, genetic algorithms requires continuous finite elementre-analysis on truss structure. BP neural network has been introduced inthis paper, and the high non-linear mapping ability of BP network has beenapplied to calculate the fitness in order to improve the convergence ofoptimization algorithms.This article embarks from the transmission tower engineeringapplications, proposes that each chord node of transmission tower is inan identical straight line called chord node straight line constraintcondition. The constraint condition could greatly decrease continuumoptimization variable(node coordinate variable)number, and theoptimization result conforms to the project reality, and enhancesoptimized efficiency.Transmission tower shape optimization model has been established inthis article. The model is for the purpose of optimizing bar cross-sectionarea and node coordinates, making the structural configuration morereasonable. The tower not only meets the requirement of stress restraint,displacement restrains, chord point straight line constraint conditionand electrical parameter, but also makes the tower lighter, reduces steel consumption and tower construction cost.A 10-bar truss and a 25-bar transmission tower shape have beenoptimized by BP-GA algorithm. The optimization results are compared withcorrelative literature, and prove the feasibility of BP-GA algorithm inpractical application. Finally an actual transmission tower is optimizedby the method in this paper. The bar cross-sectional area, the nodecoordinates, the tower root size have been optimized, which makes thetransmission tower structural configuration more reasonable. As a result,the tower weight reduces 11.4%. To be concluded, the optimized result hasbeen extremely ideal.